NVIDIA and Google Cloud announced expanded collaboration at Google Cloud Next, introducing NVIDIA Vera Rubin-powered A5X bare-metal instances capable of scaling to nearly 1 million Rubin GPUs across multisite clusters. Key announcements include Google Gemini running on Google Distributed Cloud with NVIDIA Blackwell GPUs, confidential computing support for regulated industries, NVIDIA Nemotron 3 Super integration on Gemini Enterprise Agent Platform, and a managed reinforcement learning API built on NVIDIA NeMo RL. Physical AI capabilities are extended via NVIDIA Omniverse, Isaac Sim, and Cosmos Reason 2 NIM microservices on Google Vertex AI and GKE. Real-world deployments include OpenAI running ChatGPT inference on GB300 systems and CrowdStrike using NeMo for cybersecurity AI agents.
Table of contents
Next-Generation Infrastructure: From NVIDIA Blackwell to Vera RubinSecure AI Wherever It Needs to Run: Sovereign and ConfidentialOpen Models and APIs for Agentic AIBuilding the Future of Industrial and Physical AIProven Impact: From Startups to Global EnterprisesSort: